Dutch tweaking site Tweakers.net reports AMD is developing a new threading technology for its next-generation K10 processor architecture.
AMD's "Reverse Hyper-Threading", "Anti-HT", or whatever you'd like to call it, would make it possible to run one thread on multiple cores. This is basically the opposite of HyperThreading, a technology from Intel that enables an operating system to run two threads simultaneously on one core.
It's not really clear how this technology works, the French hardware site X86-Secret heard about it from an AMD developer:
Conscious that K8 architecture could not compete with the next high-speed flagship of Intel, all its hopes is for the moment based on a new "revolutionary" technology (it is our opinion, not it his) on which AMD works in this moment for after-K8. This technology is in fact a kind of anti-HT: There or HyperThreading sought to emulate two virtual processors with a physical processor, it is a goal for AMD of emulating a single virtual processor with two (or several) physical processors.
The AMD K10 architecture will be a totally new design because the company can no longer base itself on the K8 core as it fears this old architecture won't be able to compete with Intel's new Core architecture.
Update: Someone reminded me that Intel is working on a similar technology called Mitosis. Update 2 (July 10, 2006): It looks like the Reverse Hyperthreading was just a hoax.
Use Disqus to post new comments, the old comments are listed below.
Re: AMD working on reverse Hyper-Threading technology by Anonymous on Friday, April 14 2006 @ 00:07:42 CEST
Now This is the type of technology i have been expecting for some time,
Software may not be multithreaded but if operating systems such as windows can use 3 quad core opterons as one extremely powerful processor, you wont need to completely rewrite old software.
Reply by Anonymous on Friday, April 14 2006 @ 01:12:10 CEST
this is what they needed for the cell... this solves the problem for the ghz race...
now i just wonder how efficient it will be
Reply by Anonymous on Friday, April 14 2006 @ 03:31:07 CEST
how is this different from parallel/grid/super computing?
Reply by Anonymous on Friday, April 14 2006 @ 07:52:48 CEST
Being able to load balance regular single threaded applications across multiple cores via a completely hardware-based solution, without aid from the OS, or any other software layer... this would be a lot different than your typica supercomputer cluster.